AWS Contact Center

Personalize agent assistance with Amazon Q in Connect

Imagine a world where every customer interaction is informed by a wealth of relevant data, where artificial intelligence (AI) knows your business, and where contact center agents are empowered to provide truly tailored service. As companies strive to deliver exceptional customer experiences, personalization has become a key differentiator. Customers expect tailored interactions that address their unique needs and preferences. For example, imagine you are calling to reschedule a flight due to a weather delay. In addition to booking you a seat on the next available flight, the agent would ideally know your frequent flyer status and be able to reserve you a slot in the airport lounge while you wait. However, contact centers face challenges in providing consistent, personalized service across diverse customer segments at scale. Amazon Q in Connect, our generative AI-powered assistant for contact center agents, offers personalization features that enable contact centers to deliver tailored recommendations to agents based on customer data and customizable behavior.

In this blog, we’ll dive deep into how the personalization capabilities of Amazon Q in Connect can save agents time in delivering tailored customer experiences, leading to reduced Average Handle Time (AHT) and increased First Contact Resolution (FCR). We’ll also explore how it leverages customer data to deliver context-aware recommendations, and how you can customize its behavior directly within Amazon Connect to align with your brand and industry requirements to improve both agent efficiency and end-customer experience.

Let’s explore how Amazon Q in Connect is setting a new standard for personalized customer service in the AI era.

Putting customer data to work with personalized recommendations

At the heart of the advanced capabilities of Amazon Q in Connect is the ability to provide highly personalized recommendations to agents in real-time. This transforms the way agents interact with customers by offering tailored, context-aware assistance that draws from customer data. Agents can be more present in addressing customer needs instead of searching for highly specific instructions in dense and disparate internal knowledge bases.

The journey to personalization with Amazon Q in Connect begins by integrating with multiple data sources, in addition to your knowledge base content:

  • Amazon Connect flows – This allows Amazon Q in Connect to act on contextual data, including:
    • Amazon Lex intent or conversation transcripts
    • Flow menus or contact attributes
  • Amazon Connect capabilities – By integrating with Amazon Connect Customer Profiles and Amazon Connect Cases, Amazon Q can provide continuity in issue resolution and suggest next-best actions using:
    • Customer details and preferences
    • Purchase history and product subscriptions
    • Custom attributes specific to your business, such as loyalty status or membership history
  • Third-party data sources – Amazon Q in Connect can also incorporate data from external systems, such as:
    • CRM platforms
    • Order management systems
    • Loyalty program databases

As the conversation unfolds, Amazon Q actively listens and analyzes the conversation to detect customer intent. When this contextual information is added to the session, it is combined with detected intents to pull the most relevant articles and solutions from the knowledge base and deliver it directly to the agent. A Q in Connect session is created every time a contact is associated with an Amazon Q in Connect flow block.

Let’s illustrate this with an example of an agent at a financial services company using Amazon Q in Connect. An existing customer from the US calls about their retirement plan. Amazon Q recognizes that the customer’s inquiry is related to retirement planning, and uses the associated data provided in the session from the customer profile to understand the caller’s:

  • Current retirement plan details
  • Account balances
  • Recent transactions
  • Risk tolerance assessment
  • Demographics

Based on this information, Amazon Q provides the agent with the right guidance to support the conversation based on the customer’s current retirement strategy, including tailored suggestions for plan optimization based on the customer’s age, financial goals, and relevant educational materials on retirement planning from the associated knowledge base.

If the customer wants to make changes to their plan, Amazon Q can offer the agent a customized, step-by-step guide to complete this process that takes into account:

  • The specific type of retirement account (e.g., 401(k), IRA)
  • Company policies on plan modifications
  • Regulatory requirements for retirement account changes

By providing this level of personalized, data-driven assistance, Amazon Q empowers agents to deliver expert-level service, even on complex topics like retirement planning. This improves customer experience by providing the right guidance, avoiding unnecessary escalations, and increasing agent confidence in handling diverse customer needs.

The power of these personalized recommendations lies in their ability to transform every interaction into a tailored experience, making customers feel understood and valued. As we’ll explore in the next section, the customization options for Amazon Q in Connect allow businesses to fine-tune these recommendations even further, ensuring they align with specific organizational requirements and company policies.

Customizing Amazon Q in Connect for business needs

Generative AI-powered assistance is not a one-size fits all solution, especially across different industries and company cultures. That’s why Amazon Q in Connect offers enhanced customization options, allowing you to modify the assistant’s behavior, tone, and knowledge base configurations without interacting with other services. This ensures that the assistance aligns with your company’s unique needs, industry standards, and brand voice.

To accomplish this, you can modify the Large Language Model (LLM) prompts used by Amazon Q in Connect without having to manage the LLM yourself. This allows businesses to fine-tune the behavior and responses of Q in Connect in several ways:

  1. Adjust tone and behavior:
    • Modify the communication style to match your brand voice (e.g., professional, friendly, empathetic)
    • Tailor language complexity based on your customer base (e.g., simplifying technical terms for general audiences)
    • Incorporate company-specific phrases or slogans into responses for consistency
  2. Align with brand voice and company values:
    • Ensure responses reflect your company’s mission and core values
    • Implement specific language guidelines (e.g., always addressing customers by name)
    • Customize greetings and closings to reinforce brand identity
  3. Use different content for different lines of business:
    • Configure Amazon Q to access specific knowledge bases based on the customer’s inquiry or product line to ensure agents only receive information relevant to their department
    • Designate certain responses as “fixed” for situations requiring absolute consistency (e.g., legal disclaimers, terms and conditions)

Let’s explore how a healthcare services provider might customize Amazon Q in Connect for their specific needs:

  1. Implementing empathetic tone for sensitive topics:
    • Customize prompts to use softer language when discussing serious health conditions
    • Include empathetic phrases and follow-up questions in responses related to chronic illnesses or mental health
  2. Ensuring compliance with organizational standards:
    • Set up automatic inclusion of necessary disclaimers for different types of responses
    • Configure Amazon Q to provide step-by-step guidance on compliant documentation processes
    • Maintaining confidentiality of data and HIPAA eligibility
  3. Tailoring responses to different departments:
    • Leverage separate knowledge bases for billing inquiries, claims processing, and medical consultation
    • Implement contextual access ensuring agents only receive information relevant to the contact
    • Customize language complexity based on whether the agent is speaking to a patient or a healthcare provider
  4. Providing seasonal and situational adaptability:
    • Set up temporary knowledge bases for specific health campaigns (e.g., flu season, COVID-19 updates)
    • Configure Amazon Q to proactively suggest relevant preventive care based on the customer’s profile and current health trends

For a call about a denied claim, Amazon Q might be customized to:

  1. Express empathy for the customer’s frustration
  2. Provide a clear, jargon-free explanation of the denial reason
  3. Offer a step-by-step guide for the appeal process, tailored to the specific claim type
  4. Suggest proactive measures to prevent future claim denials
  5. Include relevant disclaimers in an approved manner

This level of customization not only enhances the customer experience but also significantly improves operational efficiency. By leveraging these customization options, the organization ensures that Amazon Q in Connect becomes an extension of their brand, empowering agents to deliver consistent, compliant, and compassionate service across all customer interactions. Agents can confidently rely on the recommendations of Amazon Q, knowing they align with company policies and industry regulations. The result is a more streamlined, effective, and personalized customer service experience.

How to get started with customizing Amazon Q in Connect

To tailor Amazon Q in Connect’s behavior to your specific business needs, you can implement custom AI Prompts and AI Agents. AI Prompts are sets of instructions that guide the AI’s responses, while AI Agents combine one or more AI Prompts to create comprehensive, customized functionality. With AI Prompts, you can fine-tune the tone, language, and decision-making processes for various Amazon Q tasks such as answer generation, intent labeling, and query reformulation ensuring they align with your brand voice, industry-specific requirements, and operational procedures. AI Prompts are also where you incorporate personalization, enriching the generative AI query with customer-specific information to provide tailored responses. AI Agents, on the other hand, allow you to package these customized prompts into cohesive units that can be applied to different scenarios, such as automatic recommendations or manual searches.

These customization options can be implemented and managed through Amazon Q in Connect APIs, giving businesses full control over their customer service experience.

Supported variables for configuring AI Prompts

AI Prompts use Amazon Q in Connect system-defined data and customer-provided data as variables combinable with instructions. The following variables are supported

Variable type Variable specification in the YAML for AI Prompts Description
Q in Connect system variable {{$.transcript}} Interpolates up to three recent turns of conversation
Q in Connect system variable {{$.contentExcerpt}} Interpolates to relevant document excerpts found within the knowledge base
Q in Connect system variable {{$.query}} Interpolates to the query constructed by Q in Connect to find document excerpts within the knowledge base
Customer-provided variable {{$.Custom.<VARIABLE_NAME>}} Any customer-provided value added to a Q in Connect session

Amazon Q in Connect supports adding custom data to a session to drive the generative AI solutions presented to agents. First, add the custom data to a session using the UpdateSessionData API via the AWS Command Line Interface (CLI) as shown below, and then you can use the data added to customize AI Prompts.

aws qconnect update-session-data \
  --assistant-id <YOUR_Q_IN_CONNECT_ASSISTANT_ID> \
  --session-id <YOUR_Q_IN_CONNECT_SESSION_ID> \
  --data "[
    { \"key\": \"productId\", \"value\": { \"stringValue\": \"ABC-123\" }},
  ]"

Creating AI Prompts and AI Agents

Here’s how to create and implement custom AI Prompts:

  1. Create AI Prompts using YAML

First, you’ll need to create your AI Prompts using the YAML format. Amazon Q in Connect supports two formats: MESSAGES and TEXT_COMPLETIONS. Choose the appropriate format based on your use case:

  • MESSAGES: Use this for AI prompts that don’t require knowledge base interaction. This includes prompts that construct a query to search for knowledge base excerpts or a prompt that generates intents that are selectable by agents.
  • TEXT_COMPLETIONS: Use this for prompts that interact with a knowledge base. These prompts generate solutions using knowledge base excerpts.

Example of a MESSAGES format prompt for Query_Reformulation:

anthropic_version: bedrock-2023-05-31
system: You are an AI assistant specialized in reformulating customer service queries to improve search results.

messages:
- role: user
  content: |
    Your task is to analyze a conversation between a customer service agent and a customer, then reformulate the customer's inquiry into a     clear, concise search query. This query will be used to find relevant information in our knowledge base.

    Here is the conversation:
    <conversation>
    {{$.transcript}}
    </conversation>

    And here is the productId the customer is contacting us about

    <productId>
    {{$.Custom.productId}}
    </productId>

    Based on this conversation, create a search query that will help find the most relevant information to address the customer's needs. If a productId is specified, incorporate it in the query constructed to help scope down search results.

Output your query in the following format:

    <query> search query</query>

    Provide only the query within the tags, with no additional explanation or output.

Example of a TEXT_COMPLETIONS format prompt:

prompt: |

  You are a helpful assistant for customer service agents. Here is the query:

  <query>

  {{$.query}}

  </query>

  Here are relevant knowledge base excerpts:

  <excerpts>

  {{$.contentExcerpt}}

  </excerpts>

  Using only the data from relevant excerpts, output your answer in an <answer></answer> tag.

You can learn more about the supported prompt types and associated variables used to customize Amazon Q in Connect in the documentation.

  1. Use the CreateAIPrompt API:

Once you’ve created your YAML files, use the CreateAIPrompt API to add the prompt to Amazon Q in Connect. You can do this using the AWS CLI. Ensure you choose the correct api-format for TEXT_COMPLETIONS or MESSAGES.

After creating the AI Prompt, you will need to create a version using CreateAIPromptVersion. You will use the AI Prompt ID and the version number when associating with an AI Agent to influence Amazon Q in Connect behavior.

  1. Create an AI Agent

After creating your AI Prompts, you’ll need to create an AI Agent to combine one or more prompts. You can also use AI Agents to configure which knowledge base and associated content should be filtered. Use the CreateAIAgent API to do this:

aws qconnect create-ai-agent \

  --assistant-id <YOUR_Q_IN_CONNECT_ASSISTANT_ID> \

  --name example_answer_recommendation_agent \

  --visibility-status PUBLISHED \

  --type ANSWER_RECOMMENDATION \

  --configuration "{

    \"answerRecommendationAIAgentConfiguration\": {

      \"answerGenerationAIPromptId\": \"<AI_PROMPT_ID_WITH_VERSION_QUALIFIER>\",

      \"intentLabelingGenerationAIPromptId\": \"<AI_PROMPT_ID_WITH_VERSION_QUALIFIER>\",

      \"queryReformulationAIPromptId\": \"<AI_PROMPT_ID_WITH_VERSION_QUALIFIER>\",

      \"associationConfigurations\": [

        {

          \"associationType\": \"KNOWLEDGE_BASE\",

          \"associationId\": \"<ASSOCIATION_ID>\",

          \"associationConfigurationData\": {

            \"knowledgeBaseAssociationConfigurationData\": {

              \"overrideSearchType\": \"SEMANTIC\",

              \"maxResults\": 5,

              \"contentTagFilter\": {

                \"tagCondition\": { \"key\": \"<KEY>\", \"value\": \"<VALUE>\" }

              }

            }

          }

        }

      ]

    }

  }"

Similar to AI Prompts, you will need to create a version for you AI Agents to be used by Amazon Q in Connect. You can do this using the CreateAIAgentVersion API.

  1. Set the AI Agent for use

To apply your custom AI Agent, you can either set it as the default for your Q in Connect assistant or apply it to specific Q in Connect sessions.

To set it as the default for the assistant:

aws qconnect update-assistant-ai-agent \

  --assistant-id <YOUR_Q_IN_CONNECT_ASSISTANT_ID> \

  --ai-agent-type ANSWER_RECOMMENDATION \

  --configuration "{

    \"aiAgentId\": \"<AI_AGENT_ID_WITH_VERSION_QUALIFIER>\",

  }"

To apply it to a specific active Amazon Q in Connect session:

aws qconnect update-session \

  --assistant-id <YOUR_Q_IN_CONNECT_ASSISTANT_ID> \

  --session-id <YOUR_Q_IN_CONNECT_SESSION_ID> \

  --ai-agent-configuration "{

    \"ANSWER_RECOMMENDATION\": { \"aiAgentId\": \"<AI_AGENT_ID_WITH_VERSION_QUALIFIER>\" },

    \"MANUAL_SEARCH\": { \"aiAgentId\": \"<AI_AGENT_ID_WITH_VERSION_QUALIFIER>\" },

  }"

  1. Test and iterate

After implementing your custom prompts and agents, it’s crucial to test them thoroughly. Monitor the AI’s responses in various scenarios and gather feedback from your agents. Use this information to refine your prompts and AI agent configurations for optimal performance.

By leveraging these prompt management capabilities, you can enhance the effectiveness of Amazon Q in Connect in your contact center. Custom prompts support Amazon Q in Connect to provide responses that are not only accurate but also aligned with your specific business needs and brand voice, leading to improved agent efficiency and customer satisfaction.

 

Learn more in this video:

Conclusion

As we’ve explored throughout this post, the integration of personalized AI assistance through Amazon Q in Connect addresses the growing demand for tailored customer experiences while simultaneously enhancing operational efficiency. These capabilities offer a benefit that can significantly transform contact center operations, elevate agent performance, and enhance customer satisfaction.

At the heart of these improvements is an enhanced customer experience. By leveraging customer data, agents can provide more relevant and personalized responses, making customers feel truly understood and valued. This contextual approach not only leads to faster issue resolution but also ensures consistency across all touch points.

The impact on agent efficiency and performance is equally important. With Amazon Q in Connect, agents gain instant access to tailored information, dramatically reducing the need to search through multiple systems or knowledge bases. This real-time knowledge access not only improves the speed of service but also reduces training time for new agents. As Amazon Q provides context-aware guidance and recommendations, agents can become productive more quickly and focus their energy on handling complex issues and building stronger customer relationships.

One of the most significant benefits to businesses are the cost optimization implications. This includes the increase in First Contact Resolution (FCR) rates through understanding customer needs more quickly and accurately, in tandem with personalized recommendations increase the likelihood of resolving issues on the first contact. This context-aware assistance helps reduce Average Handle Time (AHT), while also lowering training costs, lessening the time and resources needed to train agents on complex products or procedures. Brand consistency is another crucial benefit. Customized AI Prompts can ensure that all interactions align with the company’s brand voice and values, creating a uniform brand experience. At the same time, the system allows for tailored communication styles for different customer segments or product lines, maintaining overall brand consistency while addressing diverse customer needs.

The personalization features of Amazon Q in Connect represent a significant step forward in the evolution of customer service technology. By embracing these capabilities, businesses can not only meet the current demands of their customers but also position themselves at the forefront of customer experience innovation. The future of customer service is personalized, efficient, and powered by AI – and with Amazon Q in Connect, that future is within reach today.

Resources

The journey to implementing personalized AI assistance in your contact center starts now. Here are some steps you can take:

  • Do you want to learn how to activate Amazon Q in Connect in your instance? The Amazon Connect Administrator Guide provides a comprehensive overview of Amazon Q in Connect and its features. Learn more at Enable Amazon Q in Connect for your instance.
  • Do you want to get hands-on with Amazon Q in Connect? Follow this Amazon Q in Connect workshop to learn how to activate, set up your domain, connect your content, and deliver those AI-generated responses and solutions to your agents.
  • Stay Informed: Keep an eye on the AWS Contact Center Blog for future updates and enhancements to Amazon Q in Connect and other Amazon Connect features.

Ready to transform your customer service experience with Amazon Connect? Contact us.